A5045520939- Medical Image Segmentation Techniques
- Medical Imaging Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging and Analysis
- MRI in cancer diagnosis
- AI in cancer detection
- Advanced MRI Techniques and Applications
- Spine and Intervertebral Disc Pathology
- Brain Tumor Detection and Classification
- Lung Cancer Diagnosis and Treatment
- Advanced Neural Network Applications
- Hepatocellular Carcinoma Treatment and Prognosis
- Retinal Imaging and Analysis
- Robotics and Sensor-Based Localization
- Advanced Radiotherapy Techniques
- COVID-19 diagnosis using AI
- Osteoarthritis Treatment and Mechanisms
- Domain Adaptation and Few-Shot Learning
- Face and Expression Recognition
- Ultrasound Imaging and Elastography
- Artificial Intelligence in Healthcare and Education
- Fetal and Pediatric Neurological Disorders
- Topic Modeling
- Advanced Vision and Imaging
- Advanced Image Fusion Techniques
University of Oxford
2016-2025
Open Data Institute
2019-2024
Health Data Research UK
2019-2022
American Jewish Committee
2021
Mohamed bin Zayed University of Artificial Intelligence
2021
Science Oxford
2017-2018
University of Central Lancashire
2010-2017
Institute of Biomedical Science
2016
Accurate annotation of vertebral bodies is crucial for automating the analysis spinal X-ray images. However, manual these structures a laborious and costly process due to their complex nature, including small sizes varying shapes. To address this challenge expedite process, we propose an ensemble pipeline called VertXNet. This currently combines two segmentation mechanisms, semantic using U-Net, instance Mask R-CNN, automatically segment label in lateral cervical lumbar VertXNet enhances its...
Automated analysis of structural imaging such as lung Computed Tomography (CT) plays an increasingly important role in medical applications. Despite significant progress the development image registration and segmentation methods, remain a challenging task. In this paper, we present novel approach, for which develop new mathematical formulation to jointly segment register three-dimensional CT volumes. The algorithm is based on level-set formulation, merges classic Chan-Vese with active dense...
Abstract Maturation of the human fetal brain should follow precisely scheduled structural growth and folding cerebral cortex for optimal postnatal function 1 . We present a normative digital atlas maturation based on prospective international cohort healthy pregnant women 2 , selected using World Health Organization recommendations standards 3 Their fetuses were accurately dated in first trimester, with satisfactory neurodevelopment from early pregnancy to years age 4,5 The was produced...
Purpose Compensation for respiratory motion is important during abdominal cancer treatments. In this work we report the results of 2015 MICCAI Challenge on Liver Ultrasound Tracking and extend 2D to relate them clinical relevance in form reducing treatment margins hence sparing healthy tissues, while maintaining full duty cycle. Methods We describe methodologies estimating temporally predicting liver from continuous ultrasound imaging, used ultrasound‐guided radiation therapy. Furthermore,...
In recent years, the deployment of supervised machine learning techniques for segmentation tasks has significantly increased. Nonetheless, annotation process extensive datasets remains costly, labor-intensive, and error-prone. While acquiring sufficiently large to train deep models is feasible, these often experience a distribution shift relative actual test data. This problem particularly critical in domain medical imaging, where it adversely affects efficacy automatic models. this work, we...
Foundation models are widely employed in medical image analysis, due to their high adaptability and generalizability for downstream tasks. With the increasing number of foundation being released, model selection has become an important issue. In this work, we study capabilities classification tasks by conducting a benchmark on MedMNIST dataset. Specifically, adopt various ranging from convolutional Transformer-based implement both end-to-end training linear probing all The results...
Many public PyTorch repositories implement Local Normalized Cross-Correlation Loss (LNCC) using five sequential convolution operations. This implementation is, however, slow, failing to utilize modern hardware's performance potential fully. By simply replacing these convolutions with one single group convolution, we found the training time of LNCC-based deep registration models can be halved without affecting numerical results, leading notable cost savings. We hope that this simple approach...
Abstract Introduction: Risk stratification remains a key challenge in prostate cancer (PCa) management involves risk stratification, and identification of the subgroup patients at highest progressing from localised to metastatic disease is critical. Multiparametric MRI (mpMRI) PCa diagnostic pathway. By integrating clinical parameters, mpMRI radiomics spatial transcriptomics (ST), this novel “Radio-Spatial Genomics” platform offers an exciting opportunity identify radiomic features...
Rectal tumour segmentation in dynamic contrast-enhanced MRI (DCE-MRI) is a challenging task, and an automated consistent method would be highly desirable to improve the modelling prediction of patient outcomes from tissue contrast enhancement characteristics - particularly routine clinical practice. A framework developed automate DCE-MRI segmentation, by introducing: perfusion-supervoxels over-segment classify volumes using characteristics; pieces-of-parts graphical model, which adds global...
Metastatic tumour progression is facilitated by associated macrophages (TAMs) that enforce pro-tumour mechanisms and suppress immunity. In pulmonary metastases, it unclear whether TAMs comprise tissue resident or infiltrating, recruited macrophages; the different expression patterns of these are not well established. Using mouse melanoma B16F10 model experimental metastasis, we show infiltrating (IM) change their gene from an early pro-inflammatory to a later promoting profile as lesions...
Deformable image registration, the estimation of spatial transformation between different images, is an important task in medical imaging. Deep learning techniques have been shown to perform 3D registration efficiently. However, current strategies often only focus on deformation smoothness, which leads ignorance complicated motion patterns (e.g., separate or sliding motions), especially for intersection organs. Thus, performance when dealing with discontinuous motions multiple nearby objects...
Background: Quantitative cardiovascular magnetic resonance (CMR) T1 mapping has shown promise for advanced tissue characterisation in routine clinical practise. However, is prone to motion artefacts, which affects its robustness and interpretation. Current methods correction on are model-driven with no guarantee generalisability, limiting widespread use. In contrast, emerging data-driven deep learning approaches have good performance general image registration tasks. We propose MOCOnet, a...